For historical reasons the naming of data structures and variables associated with task outputs and prerequisites is a bit confusing:

• originally task outputs were task messages, full stop
  • consequently tasks messages are the fundamental thing stored as outputs and prereqs in the code
  • and they are usually called "outputs" because they function as outputs
• then we added short names for use as triggers in the graph
  • these got variously called output "names", "labels", or "triggers", and sometimes "outputs"

So that's all a bit confusing. And now, for Cylc 8, the way we talk about optional and required outputs elevates the importance of the short names.

We should:

1. check that "output" always refers to the short name, in the docs
2. ensure that naming in the code is consistent with this
3. and possibly change the internals to store the short outputs as the fundamental entities rather than the task messages. The translation can be done first thing on receiving incoming messages